Multi-time Scale Attention Network for WEEE reverse logistics return prediction

作者：

Highlights：

• Multi-time Scale Attention Network model is proposed for WEEE return prediction.

• Data features at multi-time scales are explored in modeling temporal dependencies.

• Smooth embedding based on data aggregation is introduced to deal with high noise.

• Four kinds of time positions are used for high sensitivity to temporal dependency.

• Extensive experiments on real-world datasets demonstrate the model’s superiority.

摘要

•Multi-time Scale Attention Network model is proposed for WEEE return prediction.•Data features at multi-time scales are explored in modeling temporal dependencies.•Smooth embedding based on data aggregation is introduced to deal with high noise.•Four kinds of time positions are used for high sensitivity to temporal dependency.•Extensive experiments on real-world datasets demonstrate the model’s superiority.

论文关键词：Reverse logistics management,Waste electrical and electronic equipment,Time series forecasting,Return prediction

论文评审过程：Received 6 June 2022, Revised 7 August 2022, Accepted 15 August 2022, Available online 19 August 2022, Version of Record 27 August 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118610